NeuroFlex

Every design choice in this app is anchored to peer-reviewed evidence. Below: the studies that shaped the rules, with plain-language summaries of what they actually found.

No cognitive training app has been proven to prevent dementia. The evidence is strongest for far transfer being modest — meaning gains on trained tasks don't always generalize to daily life (Simons et al., 2016).

What does have robust evidence: aerobic exercise, sleep, the Mediterranean diet pattern, social engagement, and treating hearing loss (Livingston et al., 2020, Lancet Commission).

NeuroFlex is designed as one part of a broader cognitive-health practice — not a substitute for those bigger levers.

NeuroFlex is part of the Grand Living suite — a constellation of small, evidence-anchored wellness tools built by Noah Grand in Sandpoint, Idaho.

This app is offline-first. Everything you see runs on your device. No accounts, no servers, no tracking. Your scores live in your browser's local storage and stay there.

Five domains. Five original games. One adaptive engine. The whole thing fits in a single file.

Every game in NeuroFlex was designed from scratch around a published cognitive paradigm — never copied from a commercial app. Circuit Builder is loosely inspired by Corsi block-tapping (Corsi, 1972). Echo N-Back implements the dual N-Back working memory paradigm (Jaeggi & Au, 2008-2015). Attention Orchestra adapts the dichotic-listening tradition (Cherry, 1953). Divided Field is modeled on the Useful Field of View paradigm from the ACTIVE trial (Rebok et al., 2014). Quick Switch is a task-switching paradigm (Monsell, 2003). Lightning Match measures symbol-substitution throughput in the Digit Symbol tradition (Salthouse, 1996). Idea Chain implements Guilford's Alternate Uses Task (Guilford, 1967). The implementations are original code.

NeuroFlex scores most games with code (timing, accuracy, memory). But the Idea Chain game asks "name uses for an object" — and judging creativity needs more than counting words. So we use two different judges:

1. AI Judge (when you're signed in as a subscriber) — Claude reads your ideas, groups them by category, and rates how unusual each one is. Like having a researcher score the test.

2. Code Judge (free / signed out) — A built-in word list checks roughly which categories you covered, and a frequency table guesses how common each idea is. It's honest, but rougher — code can't really "understand" an idea the way a person (or AI) can.

Sign in with your Grand Living account to use the AI judge automatically. No API key to paste. The same subscription unlocks AI features across all Grand Living apps.

Sound effects: on